Add llms.txt generation

[fubuloubu]

Overview

The llms.txt "standard" is rapidly becoming the de-facto way that LLMs can parse a website and discover text-only information used to understand the website semantically. We should add into the Sphinx build process an aggregation/summarization pass that takes all of the markdown userguides and develops them into llms.txt/llms-full.txt files hosted under the website's /<version>/ folders. Additionally, it should also copy the .md files into .html.md files per the spec that coincide with the userguide pages (e.g. <version>/userguides/<userguide name>.html.md.) Finally, we should also take under consideration how this might play with the different versions that we host of each project's documentation sub-page, as well as how the top-level sub-page lists projects (see https://github.com/ApeWorX/apeworx.github.io for documentation portal site)

Specification

An example setup that seems to work well: https://gofastmcp.com/getting-started/welcome#text-formats

Dependencies

Rollout of sphinx-ape to all ApeWorX packages and plugins, as well as prominent 3rd-party packages (listed on our documentation portal)

[linear]

APE-1983 Add llms.txt generation

[fubuloubu]

Oh, and for products themselves, the documentation portal should link directly to the product's OpenAPI specification (e.g. /docs page) so that our products can have documentation as well